Doctor of Philosophy / This thesis is based on research on financial time series analysis using pattern recognition methods. The first part of this research focuses on univariate time series analysis using different pattern recognition methods. First, probabilities of basic patterns are used to represent the features of a section of time series. This feature can remove noise from the time series by statistical probability. It is experimentally proven that this feature is successful for pattern repeated time series. Second, a multiscale Gaussian gravity as a pattern relationship measurement which can describe the direction of the pattern relationship is introduced to pattern clustering. By searching for the Gaussian-gravity-guided nearest neighbour of each pattern, this clustering method can easily determine the boundaries of the clusters. Third, a method that unsupervised pattern classification can be transformed into multiscale supervised pattern classification by multiscale supervisory time series or multiscale filtered time series is presented. The second part of this research focuses on multivariate time series analysis using pattern recognition. A systematic method is proposed to find the independent variables of a group of share prices by time series clustering, principal component analysis, independent component analysis, and object recognition. The number of dependent variables is reduced and the multivariate time series analysis is simplified by time series clustering and principal component analysis. Independent component analysis aims to find the ideal independent variables of the group of shares. Object recognition is expected to recognize those independent variables which are similar to the independent components. This method provides a new clue to understanding the stock market and to modelling a large time series database.

Σκοπός της παρούσας διπλωματικής εργασίας είναι η μελέτη του αντίστροφου καθορισμού πηγής (inverse source localization problem) και του ρυθμού μ (mu). Έχοντας ως δεδομένο το σήμα του ΗΕΓ γίνεται προσπάθεια µέσω της εφαρμογής της μεθόδου Ανάλυσης Ανεξάρτητων συνιστωσών (ICA) να προσδιοριστούν οι συνιστώσες οι οποίες σχετίζονται με τις περιοχές του εγκεφάλου που ενεργοποιούνται από την κίνηση των χεριών. Με βάση τη λειτουργία της αισθητηριοκινητικής περιοχής του εγκεφάλου και τις ιδιότητες του ρυθμού μ, γίνεται μια μελέτη πάνω στις συνιστώσες που προκύπτουν από την ICA τόσο σε δεδομένα από πραγματική κίνηση, όσο και σε δεδομένα από νοερή κίνηση, καθώς και στην εφαρμογή που μπορεί να υπάρχει σε συστήματα Διεπαφής Εγκεφάλου – Υπολογιστή. / The subject of this diploma thesis is the study of the inverse source localization problem and the mu rhythm. Performing Independent Component Analysis (ICA) on EEG data, we try to specify the components that are related to the brain areas activated by hand movement. By focusing on the function of the somatosensory brain area and the properties or mu rhythm, we study the components resulting from Independent Component Analysis on data of both real and imaginary movement, as well as the possible implementations on Brain – Computer Interface systems.

Το θέμα της κατασκευής ενός συστήματος αναγνώρισης προτύπων για τα ηχητικά σήματα ενός ανθρώπου που κοιμάται είναι ένα από τα ανοιχτά ζητήματα της Βιοιατρικής. Στην παρούσα διπλωματική εξετάζουμε την εξαγωγή ερμηνεύσιμων σημάτων που αντιστοιχούν στον καρδιακό ρυθμό, την αναπνοή και το ροχαλητό. Χρησιμοποιούμε μεθόδους Ανάλυσης σε Ανεξάρτητες Συνιστώσες και μεθόδους Τυφλού Διαχωρισμού που εκμεταλεύονται Στατιστικές Δεύτερης Τάξης. Συμπεραίνουμε ότι οι δεύτερες είναι οι πλέον κατάλληλες όταν συνοδεύονται από ένα στάδιο προεπεξεργασίας που αφορά ανάλυση σε ζώνες συχνοτήτων. / The design of a non-intrusive Pattern Recognition System to estimate the sleep sounds is an open problem of Bioengineering. We use recordings from body-sensors to estimate the heart beat, the breathing and the snoring. In this thesis we examine the effectiveness of Independent Component Analysis for this Blind Source Separation Problem and we compare it with methods that perform Source Separation using Second Order Statistics. We take into account the temporal structure of the sources as well as the presence of noise. Our system is greatly improved by a preprocessing stage of targeted subband decomposition which uses a priori knowledge about the sources. We propose an efficient solution to this problem which is confirmed by medical data.

Σκοπός της παρούσας διπλωματικής εργασίας είναι η δημιουργία ενός συστήματος μίξης ηχητικών σημάτων και προσπάθεια διαχωρισμού τους με βάση τις μεθόδους τυφλού διαχωρισμού σημάτων. Έχοντας ως δεδομένα τα αρχικά σήματα των πηγών γίνεται προσπάθεια, αρχικά μέσω της εφαρμογής της μεθόδου Ανάλυσης Ανεξάρτητων Συνιστωσών (ICA) για την περίπτωση της στιγμιαίας μίξης και στη συνέχεια μέσω της χρήσης αλγορίθμων που στηρίζονται στο μοντέλο παράλληλου παράγοντα (PARAFAC) για την περίπτωση της συνελικτικής μίξης, να προσδιοριστούν τα σήματα των πηγών από τα σήματα μίξης. Επιπλέον, τροποποιώντας τις παραμέτρους του συστήματος που μελετάμε σε κάθε περίπτωση, προσπαθούμε να πετύχουμε τη βέλτιστη απόδοση του διαχωρισμού. / The subject of this diploma thesis is the creation of a mixing system of speech signals and the attempt of their separation using the methods of blind source separation (BSS). Considering the original source signals known, we attempt, firstly by using independent component analysis for instantaneous mixtures and then by using PARAFAC model for convolutive mixtures, to extract the original source signals from the mixing signals. Moreover, by modifying the parameters of the system we make an effort to achieve the best performance of the separation.

Cette thèse comprend deux axes de recherche. D´abord, un nouveau cadre méthodologique pour évaluer la conformité des données RSO (Radar à Synthèse d’Ouverture) multivariées à haute résolution spatiale est proposé en termes de statistique asymptotique par rapport au modèle produit. Plus précisément, la symétrie sphérique est étudiée en appliquant un test d'hypothèses sur la structure de la matrice de quadri-covariance. Deux jeux de données, simulées et réelles, sont prises en considération pour étudier la performance du test obtenu par l’analyse qualitative et quantitative des résultats. La conclusion la plus importante, en ce qui concerne la méthodologie employée dans l'analyse des données RSO multivariées, est que, selon les différents cas d’usages, une partie considérable de données hétérogènes peut ne pas s’ajuster asymptotiquement au modèle produit. Par conséquent, les algorithmes de classification et/ou détection conventionnels développés sur la base de celui-ci deviennent sub-optimaux. Cette observation met en évidence la nécessité de développer de modèles plus sophistiqués comme l'Analyse en Composantes Indépendantes, ce qui conduit à la deuxième partie de cette thèse qui consiste en l’étude du biais d’estimation des paramètres TSVM (Target Scattering Vector Model) lorsque l’ACP est utilisée. Enfin, les performances de l'algorithme sont également évaluées sous l'hypothèse du bruit gaussien corrélé spatialement. L’évaluation théorique de l'ACI comme un outil de type ICTD (In Coherent Target Decomposition) polarimétrique permet une analyse plus efficace de l’apport d’information fourni. A ce but, deux espaces de représentation sont utilisé, notamment H /alpha et TSVM / This thesis comprises two research axes. First, a new methodological framework to assess the conformity of multivariate high-resolution Synthetic Aperture Radar (SAR) data with respect to the Spherically Invariant Random Vector model in terms of asymptotic statistics is proposed. More precisely, spherical symmetry is investigated by applying statistical hypotheses testing on the structure of the quadricovariance matrix. Both simulated and real data are taken into consideration to investigate the performance of the derived test by a detailed qualitative and quantitative analysis. The most important conclusion drawn, regarding the methodology employed in analysing SAR data, is that, depending on the scenario under study, a considerable portion of high heterogeneous data may not fit the aforementioned model. Therefore, traditional detection and classification algorithms developed based on the latter become sub-optimal when applied in such kind of regions. This assertion highlights for the need of the development of model independent algorithms, like the Independent Component Analysis, what leads to the second part of the thesis. A Monte Carlo approach is performed in order to investigate the bias in estimating the Touzi's Target Scattering Vector Model (TSVM) parameters when ICA is employed using a sliding window approach under different scenarios. Finally, the performance of the algorithm is also evaluated under Gaussian clutter assumption and when spatial correlation is introduced in the model. These theoretical assessment of ICA based ICTD enables a more efficient analysis of the potential new information provided by the ICA based ICTD. Both Touzi TSVM as well as Cloude and Pottier H/alpha feature space are then taken into consideration for that purpose. The combined use of ICA and Touzi TSVM is straightforward, indicating new, but not groundbreaking information, when compared to the Eigenvector approach. Nevertheless, the analysis of the combined use of ICA and Cloude and Pottier H/alpha feature space revealed a potential aspect of the Independent Component Analysis based ICTD, which can not be matched by the Eigenvector approach. ICA does not introduce any unfeasible region in the H/alpha plane, increasing the range of possible natural phenomenons depicted in the aforementioned feature space.

Nas últimas duas décadas, estudos empíricos em finanças têm utilizado o método de estudo de eventos para detectar retornos anormais no entorno de eventos que, teoricamente, deveriam ser incorporados instantaneamente no preço dos títulos. O método de estudo de eventos, a partir da década de 90, com a massificação das planilhas eletrônicas e dos pacotes estatísticos, se popularizou no meio acadêmico brasileiro, sendo um dos principais métodos de pesquisa em finanças com ênfase em mercado de capitais ou finanças corporativas. Apesar da eficácia do método em detectar a anormalidade dos retornos, comprovada em diversos estudos empíricos, acredita-se que o método seja pouco eficiente em determinar a verdadeira amplitude do retorno anormal, uma vez que são necessários pressupostos estatísticos e argumentos econômico-financeiros que podem não ser válidos. O fato de que cada modelo apresenta um desempenho diferente de captura dos retornos anormais contribui com a tese de que os modelos utilizados atualmente não conseguem filtrar totalmente o retorno anormal da série normal. Portanto, este estudo teve como objetivo principal testar a aplicabilidade do método de Análise de Componentes Independentes - ICA - em detectar retornos anormais em séries temporais e comparar o seu desempenho com os modelos geradores de retornos anormais mais utilizados em testes empíricos. Com este objetivo, foram realizadas milhares de simulações envolvendo parâmetros semelhantes aos do mercado de ações brasileiro, com o uso de algoritmos de simulação elaborados exclusivamente para esta finalidade. Os resultados sugerem que o método ICA é capaz de detectar anormalidades em séries temporais, fornecendo, desta forma, a descoberta do real impacto do retorno anormal nos elementos da amostra, necessitando apenas de uma modelagem prévia em função do tamanho da amostra e sua variância. / In the last two decades financial empiric studies have used the event study method to detect abnormal return on events that in theory should be instantly incorporated on securities price. This method became popular to Brazilian academic environment through the intensification usage of electronic worksheet and statistic packages in the 90`s turning into one of the main research methods for financial studies with emphasis on stock market and corporative financing. Despite the efficiency of the method in detecting abnormalities it`s believed that it`s least effective on establishing the real amplitude of the abnormal return considering that statistics presupposed and economic and financial arguments may not be valid. The fact that each model shows a different performance on capturing abnormal returns contributes to the idea that today`s models can`t completely filter the abnormal return on a normal series. Therefore this study has as a main objective to test the applicability of the Independent Component Analysis method – ICA – in detecting abnormal returns in time series and comparing its performance against abnormal return generating models more used on empiric tests. With this objective, thousands of simulations involving parameters similar to the Brazilian stock market with the usage of simulation algorisms elaborated exclusively for this purpose. The results suggest that ICA method is capable of detecting abnormalities in time series supplying in this form a discovery on the real impact of abnormal return on sample elements needing only a previous molding due to the size of its sample and variance.

Submitted by Pablo F. Oliveira (pablo.perque@gmail.com) on 2012-02-29T12:36:59Z No. of bitstreams: 1 Dissertacao - Pablo Frisanco Oliveira -final.pdf: 1039827 bytes, checksum: 63d8e3ff3c6593d9ef449829e78e77c1 (MD5) / Approved for entry into archive by Gisele Isaura Hannickel (gisele.hannickel@fgv.br) on 2012-02-29T12:45:22Z (GMT) No. of bitstreams: 1 Dissertacao - Pablo Frisanco Oliveira -final.pdf: 1039827 bytes, checksum: 63d8e3ff3c6593d9ef449829e78e77c1 (MD5) / Made available in DSpace on 2012-02-29T12:55:43Z (GMT). No. of bitstreams: 1 Dissertacao - Pablo Frisanco Oliveira -final.pdf: 1039827 bytes, checksum: 63d8e3ff3c6593d9ef449829e78e77c1 (MD5) Previous issue date: 2012-01-30 / The aim of the present work is verify if, when the higher moments (skewness and kurtosis) are taken in consideration for carry trade portfolio allocation optimization, an investor can be better off than the traditional allocation, which prioritizes only the first two moments (mean and variance). The hypothesis of the research is that a carry trade currency exhibits non-Normal returns distribution, and its higher moments have a dynamic which can be modeled by GARCH-type model, in this specific case IC-GARCHSK. This model consists of one equation to each of the independent components’ conditional moments, named the returns, variance, the skewness, and the kurtosis. Another hypothesis is that a CARA (constant absolute risk aversion) utility function investor can have its function approximated by 4th order Taylor expansion. The work’s strategy is modelling the dynamics of the daily log-returns series’ moments of some carry trade currencies using the model above and dynamically estimate the optimal allocation which maximizes the investor’s expected utility function. The results show that the investor can benefit from taking in consideration the series’ higher moments, once this portfolio exhibited smaller opportunity cost than one that uses only mean and variance as criteria. / O objetivo do presente trabalho é verificar se, ao levar-se em consideração momentos de ordem superior (assimetria e curtose) na alocação de uma carteira de carry trade, há ganhos em relação à alocação tradicional que prioriza somente os dois primeiros momentos (média e variância). A hipótese da pesquisa é que moedas de carry trade apresentam retornos com distribuição não-Normal, e os momentos de ordem superior desta têm uma dinâmica, a qual pode ser modelada através de um modelo da família GARCH, neste caso IC-GARCHSK. Este modelo consiste em uma equação para cada momento condicional dos componentes independentes, explicitamente: o retorno, a variância, a assimetria, e a curtose. Outra hipótese é que um investidor com uma função utilidade do tipo CARA (constant absolute risk aversion), pode tê-la aproximada por uma expansão de Taylor de 4ª ordem. A estratégia do trabalho é modelar a dinâmica dos momentos da série dos logartimos neperianos dos retornos diários de algumas moedas de carry trade através do modelo IC-GARCHSK, e estimar a alocação ótima da carteira dinamicamente, de tal forma que se maximize a função utilidade do investidor. Os resultados mostram que há ganhos sim, ao levar-se em consideração os momentos de ordem superior, uma vez que o custo de oportunidade desta foi menor que o de uma carteira construída somente utilizando como critérios média e variância.

Orientador: Prof. Dr. Ricardo Suyama / Dissertação (mestrado) - Universidade Federal do ABC, Programa de Pós-Graduação em Engenharia da Informação, 2016. / O problema de separação cega de fontes no contexto de misturas subparametrizadas tem sido investigado por meio de abordagens que exploram diferentes características dos sinais de interesse, dentre as quais se destaca a esparsidade. Mesmo que os sinais de interesse não sejam originalmente esparsos, é possível, em algumas aplicações, que estes sejam modificados a fim de apresentar um maior grau de esparsidade, e assim facilitar o processo de separação dos sinais. No presente trabalho, comparamos o desempenho de diferentes técnicas de estimação dos sinais de áudio, no contexto de misturas sub-parametrizadas, considerando que os mesmos tenham sido esparsificados antes do processo de mistura. Os resultados obtidos estendem análises preliminares realizadas, e indicam que este préprocessamento traz ganhos efetivos para o processo de estimação dos sinais. Além disso, distintos estudos foram unificados e uma nova proposta estabelecida, obtendo-se um resultado de estimação considerável no âmbito perceptual, segundo análises realizadas. / The blind source separation problem have been investigated through approaches that explore specific characteristics of the signals of interest, among which stands out the sparsity. Even if the original sources aren¿t sparse, some modifications can be done to rebuild signal with greater degree of sparsity, making the separation process easier. In this work we compare the performance of different estimation methods of audio signals, in the underdetermined context, considering that they have been sparsified before the mixing process. The results extend preliminary studies and show that this process may increase the performance of the estimation process. In addition to that, different studies were merged and a new proposal was established, which results are remarkable according to the perceptual analysis.

Orientador: Prof. Dr. Francisco José Fraga da Silva / Dissertação (mestrado) - Universidade Federal do ABC, Programa de Pós-Graduação em Engenharia da Informação, 2016. / Inibição pré-pulso (IPP) consiste numa redução na magnitude do reflexo de sobressalto a um estímulo auditivo forte (pulso), quando este é precedido por um fraco (pré-pulso). Em humanos, a IPP, geralmente, é medida por eletromiografia (EMG),. Este comportamento tem sido extensivamente investigado em estudos relacionados à esquizofrenia, uma vez que este déficit sensório-motor desempenha um papel central na sua fisiopatologia. No entanto, os mesmos estímulos auditivos que provocam o reflexo de sobressalto acústico também provocam intensas respostas auditivas evocadas, que podem ser medidas por eletroencefalografia (EEG). Analisar estes dois tipos de resposta adquiridos simultaneamente é uma grande oportunidade para investigar a dependência e interdependência de suas vias neurais. Este trabalho teve como objetivo registrar e analisar simultaneamente o reflexo de sobressalto acústico (usando EMG) e as respostas auditivas evocadas (utilizando EEG), para caracterizar o fenômeno de IPP em voluntários saudáveis, visando futura aplicação em pacientes esquizofrênicos. Utilizando técnicas avançadas de remoção de artefatos por meio da Análise de Componentes Independentes, verificou-se que após a remoção de artefatos houve uma melhor caracterização da IPP usando EEG. Especificamente, houve um aumento substancial na quantidade de diferenças estatisticamente significativas entre as respostas auditivas evocadas (medidas em vários eletrodos) com e sem pré-pulso, em comparação com os dados de EEG brutos (sem remoção de artefatos), bem como um aumento no número e na intensidade de correlações estatisticamente significativas entre o reflexo de sobressalto acústico (usando EMG) e as respostas auditivas evocadas, medidas em diversos eletrodos do EEG. Tais resultados promissores são motivadores no sentido de se tentar aplicar as mesmas técnicas em um próximo estudo envolvendo pacientes esquizofrênicos. / Prepulse inhibition (PPI) is a reduction in the magnitude of the acoustic startle reflex when a strong auditory stimulus (pulse) is preceded by a weaker one (prepulse). In humans, PPI is usually measured by electromyography (EMG). This behavior has been extensively minvestigated in studies related to schizophrenia, since the sensorimotor deficit plays a central role in its physiopathology. However, the same auditory stimuli that produce the acoustic startle reflex also trigger strong auditory evoked responses, which can be measured by electroencephalography (EEG). Analyzing these two response types acquired simultaneously is a great opportunity to investigate the dependence and interdependence of their neural pathways. This study intended to record and analyze both the acoustic startle reflex (using EMG) and the auditory evoked responses (using EEG), to characterize PPI in healthy volunteers, aiming future application in schizophrenic patients. Using advanced artifact removal techniques by means of Independent Component Analysis, it was found that after artifact removal there was a better characterization of PPI using EEG. Specifically, there was a substantial increase in the number of statistically significant differences between the auditory evoked responses (measured on multiple electrodes) with and without pre-pulse when compared to raw EEG data (without removing artifacts), as well as an increase in the number and intensity of statistically significant correlations between the acoustic startle reflex (using EMG) and the auditory evoked responses, measured in several EEG electrodes. These promising results are encouraging in order to try to apply the same techniques in a following study of schizophrenic patients.

Nas últimas duas décadas, estudos empíricos em finanças têm utilizado o método de estudo de eventos para detectar retornos anormais no entorno de eventos que, teoricamente, deveriam ser incorporados instantaneamente no preço dos títulos. O método de estudo de eventos, a partir da década de 90, com a massificação das planilhas eletrônicas e dos pacotes estatísticos, se popularizou no meio acadêmico brasileiro, sendo um dos principais métodos de pesquisa em finanças com ênfase em mercado de capitais ou finanças corporativas. Apesar da eficácia do método em detectar a anormalidade dos retornos, comprovada em diversos estudos empíricos, acredita-se que o método seja pouco eficiente em determinar a verdadeira amplitude do retorno anormal, uma vez que são necessários pressupostos estatísticos e argumentos econômico-financeiros que podem não ser válidos. O fato de que cada modelo apresenta um desempenho diferente de captura dos retornos anormais contribui com a tese de que os modelos utilizados atualmente não conseguem filtrar totalmente o retorno anormal da série normal. Portanto, este estudo teve como objetivo principal testar a aplicabilidade do método de Análise de Componentes Independentes - ICA - em detectar retornos anormais em séries temporais e comparar o seu desempenho com os modelos geradores de retornos anormais mais utilizados em testes empíricos. Com este objetivo, foram realizadas milhares de simulações envolvendo parâmetros semelhantes aos do mercado de ações brasileiro, com o uso de algoritmos de simulação elaborados exclusivamente para esta finalidade. Os resultados sugerem que o método ICA é capaz de detectar anormalidades em séries temporais, fornecendo, desta forma, a descoberta do real impacto do retorno anormal nos elementos da amostra, necessitando apenas de uma modelagem prévia em função do tamanho da amostra e sua variância. / In the last two decades financial empiric studies have used the event study method to detect abnormal return on events that in theory should be instantly incorporated on securities price. This method became popular to Brazilian academic environment through the intensification usage of electronic worksheet and statistic packages in the 90`s turning into one of the main research methods for financial studies with emphasis on stock market and corporative financing. Despite the efficiency of the method in detecting abnormalities it`s believed that it`s least effective on establishing the real amplitude of the abnormal return considering that statistics presupposed and economic and financial arguments may not be valid. The fact that each model shows a different performance on capturing abnormal returns contributes to the idea that today`s models can`t completely filter the abnormal return on a normal series. Therefore this study has as a main objective to test the applicability of the Independent Component Analysis method – ICA – in detecting abnormal returns in time series and comparing its performance against abnormal return generating models more used on empiric tests. With this objective, thousands of simulations involving parameters similar to the Brazilian stock market with the usage of simulation algorisms elaborated exclusively for this purpose. The results suggest that ICA method is capable of detecting abnormalities in time series supplying in this form a discovery on the real impact of abnormal return on sample elements needing only a previous molding due to the size of its sample and variance.